PEDS Advance Access originally published online on July 20, 2007
Protein Engineering Design and Selection 2007 20(8):397-403; doi:10.1093/protein/gzm033
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Crystal structures of high affinity human T-cell receptors bound to peptide major histocompatibility complex reveal native diagonal binding geometry
1 Avidex Limited (subsidiary of Medigene Ag), 57c Milton Park, Abingdon, Oxon OX14 4RX, UK 2 DARTS, CCLRC Daresbury Laboratory, Warrington, Cheshire WA4 4AD, UK 3Center for Molecular Immunology, Institute of Microbiology, Chinese Academy of Sciences, Beijing 100080, P. R. China 4Medical Biochemistry and Immunology, Henry Wellcome Building, University of Cardiff School of Medicine, Heath Park, Cardiff CF14 4XN, UK
5 To whom correspondence should be addressed. E-mail: bent.jakobsen{at}avidex.com
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Naturally selected T-cell receptors (TCRs) are characterised by low binding affinities, typically in the range 1–100 µM. Crystal structures of syngeneic TCRs bound to peptide major histocompatibility complex (pMHC) antigens exhibit a conserved mode of binding characterised by a distinct diagonal binding geometry, with poor shape complementarity (SC) between receptor and ligand. Here, we report the structures of three in vitro affinity enhanced TCRs that recognise the pMHC tumour epitope NY-ESO157–165 (SLLMWITQC). These crystal structures reveal that the docking mode for the high affinity TCRs is identical to that reported for the parental wild-type TCR, with only subtle changes in the mutated complementarity determining regions (CDRs) that form contacts with pMHC; both CDR2 and CDR3 mutations act synergistically to improve the overall affinity. Comparison of free and bound TCR structures for both wild-type and a CDR3 mutant reveal an induced fit mechanism arising from restructuring of CDR3 loops which allows better peptide binding. Overall, an increased interface area, improved SC and additional H-bonding interactions are observed, accounting for the increase in affinity. Most notably, there is a marked increase in the SC for the central methionine and tryptophan peptide motif over the native TCR.
Keywords: Biacore analysis/class I pMHC/crystal structure/high affinity TCR
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The adaptive immune system employs two principally different defence mechanisms in response to pathogens. Humoral responses are characterised by the production of antibodies that react with (usually) unprocessed antigens, mostly glycoproteins and polysaccharides. Cell mediated immune responses involve T-cell receptors (TCRs) that recognise proteolytic fragments from antigenic proteins in the context of class I or class II major histocompatibility complex (MHC) on target cells. However, unlike antibodies, TCRs do not undergo a process of somatic hypermutation and affinity maturation that generate high affinity clonal populations. Rather, T-cells are subject to positive and negative selection such that only those expressing TCRs with low affinities (Kd = 1–100 µM) for pMHC survive (van der Merwe and Davis 2003
). Therefore, most naturally occurring T-cell receptors display low affinities for their target molecules (Davis et al., 1998; Willcox et al., 1999a).
Each TCR is a hetero-dimer composed of either 
and ß chains, or 
and 
chains. The available TCR/pMHC complex structures have provided an insight into the molecular details of MHC restriction and recognition by TCRs (Rudolph et al., 2006). Methods to produce soluble TCRs (Willcox et al., 1999b; Boulter et al., 2003) combined with phage display technology have recently enabled us to produce high affinity versions of soluble human TCRs. These reagents may have broad applications as targeting molecules in biological assays and as therapeutics (Ashfield and Jakobsen, 2006; Molloy et al., 2005).
Both class I and class II MHC molecules are hetero-dimers with similar overall architecture composed of three domains, one -helix/ ß-sheet (ß) superdomain that forms the peptide binding site and two IgG-like domains. In class I MHC, the peptide binding site is constructed as a groove formed by the 1 and 2 helices of the heavy chain. Class I MHC molecules usually bind peptides of 8–10 residues in length (averaging 9 residues, termed P1–P9) in an extended conformation with anchor residues buried into specificity pockets, which vary from allele to allele (Rudolph et al., 2006). This binding mode leaves the central peptide side chains available for interaction with the TCR. Longer peptides are accommodated by extension at the C-terminus, or by the bulging out of the binding groove (Stern and Wiley, 1994; Tynan et al., 2005). Unlike class I MHC, the peptide binding groove of class II MHC is formed from two non-covalently linked subunits of and ß-chains; the more open peptide binding groove of class II MHC allows the accommodation of longer peptides.
The comparison of previously reported TCR/pMHC crystal structures reveals a flexible diagonal docking mechanism whereby the TCR crosses the long axis of the MHC-peptide binding groove at an angle and with the V domain poised above the N-terminal half of the peptide (Rudolph et al., 2006). The complementarity determining region (CDR) loops that form peptide contacts show a preference for centrally located peptide positions. Comparisons of crystal structures of free and bound TCR show that the majority of CDR conformational changes are limited to the hypervariable CDR3
or CDR3
loops. However, in one TCR (LC13) complex with MHC (TCR/MHC), both CDR3 loops as well as CDR1 and CDR2, underwent conformational change (Kjer-Nielsen et al., 2003). In this structure, there is a large lateral shift of the LC13 TCR towards the C-terminus which forms dominant contact area with P7 Tyr rather than the more common P5 peptide side chain. Interestingly, the mutation of P7 Tyr to Phe reduces affinity by a factor of 10 suggesting that dominant interactions are formed at this residue. During docking, the buried surface area ranges between 600 and 2000 Å2 and the majority of this result from MHC contact with the TCR (Rudolph et al., 2006). The shape complementarity (SC), which is a measure of geometrical match between protein interfaces, ranges from 0.41 to 0.75.
The directed evolution of high affinity 1G4 TCR that recognises NY-ESO-1157–165 peptide associated MHC, in the context of human leukocyte antigen (pHLA), has been described previously (Li et al., 2005). Although the structural basis of TCR peptide specificity has been well documented (Rudolph et al., 2006), it is difficult to extrapolate this data to explain how high affinity is generated in the case of the 1G4 TCR. We have solved the crystal structures of ‘free’ and ‘bound’ 1G4 TCR with mutations in the CDR3 loops that give moderate improvements in affinity and 1G4 TCR with mutations in both CDR2 and CDR3 loops that give dramatic improvements in affinity. From these structures, it is apparent that high affinity arises through an improved TCR/pMHC surface complementarity resulting from the restructuring of the CDR2/CDR3 loops around the peptide.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Molecular biology, expression and refolding
Peptide-HLA complexes were prepared as previously described (Garboczi et al., 1992) by expressing HLA-A*0201 heavy chain truncated at residue Pro-276 and ß-2 microglobulin separately in Escherichia coli in the form of inclusion bodies, followed by in vitro refolding with synthetic peptide NY-ESO157–165 (SLLMWITQC). pHLA for Biacore analysis was prepared similarly, but with the HLA fused to a biotinylation tag which was biotinylated in vitro by BirA enzyme (O'Callaghan et al., 1999).
High affinity 1G4 TCR, which recognises HLA-A2 NY-ESO157–165, were generated using phage display as previously described (Li et al., 2005). Plasmids containing the target genes were expressed in BL21(DE3) E.coli cells in the form of inclusion bodies and refolded to give soluble protein as previously described (Boulter et al., 2003). The protein was buffer exchanged into 10 mM Hepes buffer pH7, 150 mM NaCl, 3 mM EDTA for Biacore studies, or 10 mM NaCl, 10 mM Tris pH8 for crystallisation.
Binding analysis by BiacoreTM surface plasmon resonance
Binding analysis was performed on a BiacoreTM 3000 machine using CM-5 (research grade) chip and HBS (Biacore) buffer. Streptavidin was immobilized on all flow cells using amine coupling to a level of > 1000 RU (response units). Biotin tagged peptide-HLA (pHLA) was flowed over the streptavidin coated surface at a concentration of 
10 µg/ml until 150 RU pHLA was bound. Control surfaces were coated with non-binding pHLA-A2flu (influenza matrix peptide), pHLA-A2Tax (HTLV-1 tax11–19 peptide) or were left simply coated with streptavidin.
Equilibrium binding was carried out on low affinity mutants by injecting 1–200 µM concentration of TCR over the chip as previously described (Li et al., 2005). Kinetic binding data were generated using the KINJECT program to inject 10–100 nM TCR over the flow cells at 50 µl/min. Data were collected for at least 1 h and kinetic constants were derived using the curve-fitting facility of the BIAevaluation program (version 3.0, Biacore) that deploys the Marquardt-Levenberg algorithm. Rate equations were derived from the simple 1:1 Langmuir binding model (A + B 
AB). Other curve fitting was performed in Origin version 3 (MicroCal).
Crystallisation and data collection
1G4 TCR/HLA-A2 NY-ESO157–165 complexes were mixed at a 1:1 mole ratio (10 mg/ml final protein concentration) and purified by SEC (S75 HR column) into 10 mM NaCl, 10 mM Tris pH8. Crystals were grown using the hanging drop method at 18°C using Hampton Screen Cryo Buffer 41 (85 mM Na HEPES buffer pH7.5, 8.5% iso-propanol, 17% PEG 4000, 15% glycerol). After optimisation of initial crystallisation conditions, crystals appeared as thin plates.
Single crystals of the different samples, measuring up to 200 x 200 x 100 µm, were mounted in thin nylon loops and cryo-cooled in liquid nitrogen. Samples were irradiated at SRS Stations 14.1 or 14.2 (Daresbury laboratory, Warrington, UK). Each crystal was set in a random orientation and a complete data set collected on an ADSC Q4 CCD detector, with the rotation method, at an exposure rate of 30 to 90 s per 1° rotation. Diffraction maximum resolution ranged between 1.9Å and 2.4Å. The relevant statistics are included in Table II. MOSFLM (Leslie, 1992) was used to estimate structure factor intensities, SCALA (Evans, 1993) for merging and reducing the data sets to the unique indices, and TRUNCATE (French and Wilson, 1978) for producing amplitudes and other statistics (CCP4 program suite; CCP4 1994).
|
Structure solution and refinement
The structures were solved with molecular replacement methods using AMoRe (Navaza, 2001) or MOLREP (Vagin and Teplyakov, 1997). The search probe was taken from PDB entry 2F53, the model of NYEc49c50. Satisfactory solutions were subjected to rigid-body refinement, followed by rounds of restrained positional and thermal parameters refinement with REFMAC5 (Murshudov et al., 1997), and model adjustment in graphics sessions using O (Jones et al., 1991) or COOT (Emsley and Cowtan, 2004). The resulting models had average geometrical deviations from ideal values within the expected range. Completion of the structural analysis was performed with the CCP4 suite. Figures were made with PYMOL (Delano, 2002).
The refined models have been submitted to the RCSB Protein Data Bank under accession codes 2P5E, 2P5W, 2PYE and 2PYF, respectively.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Biacore binding analysis
Kinetic analysis of all high affinity 1G4 TCR mutants showed that the improved affinity results mainly from a decreased off-rate (Table I). This is exemplified by the highest affinity mutant c58c61 (KD = 48 pM) which, containing a combination of alpha and beta CDR2/CDR3 mutations, produced a 4800-fold decreased off-rate compared with the wild-type TCR. In contrast, the on-rate appears to have increased moderately by 143-fold for this mutant.
|
The c58c62 mutant was designed to retain most of the CDR loop mutations from high affinity c58c61, apart from its CDR2ß loop, which was identical to the c49c50 mutant (Table I). The resultant c58c62 mutant displays a decreased off-rate of 4000 and extended T1/2 values of 450 min compared with wild-type TCR (Table I).
The mutation of CDR2 (c49c50) loops alone results in 550 fold decreased off-rate and a moderate (45) fold increase in on-rate, suggesting that the improved affinity of c49c50 mutant results mainly from a decreased off-rate.
Overall, the improvement in affinity gained by the mutant TCRs correlated well with improved half-lives of the mutant complexes. The highest affinity mutant TCR showed a 4500-fold improvement in T1/2 values.
To investigate the role of TrpP5 as proposed contact ‘hot spot’, we have made a c58c61 and wild-type 1G4 TCR with the point mutation Y31D. Introduction of this mutation, designed to knock-out 
stacking interactions, caused a dramatic reduction in affinity in both TCRs. The affinity of wild-type Y31D clone was reduced by nearly 10-fold and hence was below the levels of accurate measurement by Biacore. The mutant protein c58 (Y31D)/c61 showed 3.8 x 105 fold drop in affinity, thus demonstrating the importance of this contact (Table I).
The crystallographic analysis produced well-behaved models, with deviations from ideal values of bond lengths and angles within the expected range (Table II). The general features of the fold were well preserved across all the models. Differences were concentrated at the CDR2/CDR3 mutation sites. Superposition statistics are included in supplementary materials.
As discovered earlier, the mutant TCR/pMHC complexes had different interface geometries from those found in the higher resolution model of the wild-type complex, 2BNR (Chen et al., 2005). However, the lower resolution model, 2F54 (Dunn et al., 2006), had two copies in the asymmetric unit, one with interface geometry similar to that of 2BNR and the other much closer to that of the c49c50 mutant, 2F53 (Dunn et al., 2006). This observation was also true in the case of the c5c1 (2PYE), c58c61 (2P5E) and c58c62 (2P5W) mutant models. Figure 1A shows a multiple superposition of all the mutants onto the coordinates of wt-AV2, the copy of the wild-type complex in 2F54 with the closest interface geometry. Previous studies of TCR-pMHC complexes compared the superposition based on the coordinates of the MHC 1/2 domain, the groove which binds the peptide (Dunn et al., 2006), but in this work, the superposition was based on the coordinates of the peptide, to avoid the variable geometries mentioned above. Figure 1B shows the multiple superposition of the peptide from all the mutant complexes onto the co-ordinates of wt-AV2. The well-matched superposition of peptides suggests that this binding surface remains invariant across the mutants.
|
Calculations of interface areas between the TCR and pMHC parts of the complex were completed with the Protein Interfaces, Surfaces and Assemblies server at the EBI (http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html). Bearing in mind that the interface geometry should be compared with wt-AV2, the interface area increases significantly when mutations occur simultaneously in both CDR2 and CDR3 (Table III). The energy of interaction across CDR2/3-pMHC interfaces may be quantified through approximations that exclude solvent molecules from calculations, since solvent molecules could not be located in the relatively low resolution (2.7
) of the wt-AV2 model. This space was eventually filled by the mutant side chains.
|
The calculated SC index (Lawrence and Colman, 1993
), listed in Table IV, showed three notable trends associated with higher affinity: (i) rising SC index of TCR/pMHC; (ii) rising SC index of TCR/peptide; (iii) rising SC index of TCR/M4W5 peptide motif. It is likely that these trends act cooperatively and produce the substantial increase in the measured affinity. The overall SC index between the surfaces of pMHC and TCR at the interface increased in all cases, from 0.71 to 0.77 and correlates well with the increased affinity. It should be noted that this increase was discernable only when CDR2/ß loops, or CDR3/ß loops, were mutated. Likewise, the SC index of TCR/pMHC in all mutant TCRs was found to increase compared with the wild-type native protein but appears to be indifferent to the type of mutation. The SC index between TCR/peptide did not change dramatically and remained between 0.83 and 0.84 irrespective of the mutation. Only minor changes occurred for CDR3 mutations alone (c5c1), or CDR2 mutations alone (c49c50). However, when both CDR2 and CDR3 loops were mutated, as in the case of c58c61, or c58c62, the SC index increased to 0.87, suggesting an improved TCR/peptide contact has been made. The SC index for the TCR/M4W5 peptide motif followed a similar trend and appeared to increase exclusively in the combined CDR
/CDR3 mutants, thus suggesting that an improved contact has been made.
|
Comparison of the CDR3 loops
The effect of the CDR3 mutations can be traced systematically via comparisons of the free wild-type (2BNU) and c5c1 mutant TCRs, free c5c1 mutant TCR and its complex with pMHC, c49c50 (CDR2 only) mutant complex (2F53) and c5c1 (CDR3 only) mutant complex, and the combined c58c61 and c58c62 (CDR2/CDR3) mutant complexes. The CDR3 loops of the TCR show significant conformational flexibility as seen in unbound and bound c5c1 (Fig. 2), and bound wild-type and c58c61 (Fig. 3).
|
|
Comparison between free and bound c5c1 TCR structures clearly shows binding with an ‘induced fit mechanism’. The insertion of the central M4W5 core peptide into the large CDR3
arc opens it up significantly, such that the aromatic ring of Tyr100 can engage with a reoriented Leu94ß (mutated from Val) and complete a ‘lock’ around the M4W5 motif (Fig. 2). A much more dramatic movement occurs at Asp97 at the apex of the large arc, which curls upward and outward, making a side chain contact with Thr99 via ionic contacts. While the interaction between Asp97 and Thr99 could be stable in a solution of free c5c1 TCR, the crystal packing might have forced it to adopt an elongated form. In contrast, the side chain of Tyr100 does not appear to move significantly upon complex formation, this therefore requires a longer side chain in CDR3ß to reach across and complete this locking action, hence the mutation of Val94Leu allows an improved contact. The remaining point mutations in the CDR3 loop, Leu94 and Leu95, appear to be unfavourable in the free TCR as they are accessible to the solvent; however, when the complex is formed they make favourable approaches to Ser51 in CDR2 and Ala30 in CDR1thereby enhancing the stability of the new conformation (not shown).
Figure 3 shows unambiguously how the CDR3 (c58c61) mutations can be separated and their effect can be made cooperative. The agreement between wild type and c58c61 in the CDR3 loops is about the same for the common atoms, at an rmsd of 0.8 . However, the new side chain atoms in c58c61 stabilise the CDR3 loops by forming a ‘lock’ around core peptide M4W5 motif during MHC binding, as observed for c5c1. These mutations emphasise the rigid nature of the c58c61 mutant conformation, unlike the flexible loops observed in the wild-type TCR (Fig. 3). Figure 4 shows clearly that all CDR3 mutants (c5c1, c58c61 and c58c62) adopt very closely the same rigid CDR3 ‘lock’. Together with the calculated SC index data, the CDR2/CDR3 
/ß combinations appear to be a perfect match for the peptide M4W5 motif.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
This is the first definitive report describing the structure-affinity correlation of high affinity human TCR/ pMHC complexes obtained by the in vitro mutation of CDR loops. The key findings from the present data suggest that the characteristic diagonal docking of high affinity 1G4 TCR remains unaffected and two key peptide residues (MetP4 and TrpP5) on the pMHC form the predominant contact (‘hot spot’) with the mutated CDR3 loops. This hypothesis was tested by designing Y31D
knock-out wild-type and c58c61 mutants, i.e. designed to remove the ring stacking contacts between CDR1/3 loops and peptide TrpP5 residue observed in all wild-type and mutant structures. The dramatic reduction of binding confirmed our hypothesis that this interaction is essential and invariant in both wild type and high affinity TCR/pMHC complexes. Further supporting evidence for this was obtained from peptide mutation studies that removed specific CDR3 loop contacts with MetP4 and TrpP5 residues; the Ala substitution of MetP4 or TrpP5 residues was found to abolish binding to all mutants and wild-type TCRs (data not shown).
To date, there is no reported structural data of high affinity TCRs for direct comparison with the present study since most naturally selected TCRs have low affinities (1–100 mM); therefore, a structural comparison was made with high affinity Fab/pHLA A1:MAGE-1 (Hulsmeyer et al., 2004). Affinity-matured Fab-Hayb3 have indicated that burial of increasing amount of apolar surfaces, in particular at the interface periphery, and improved SC are the key structural elements that contribute to high affinity. The crystal structure of Fab/pHLA A1:MAGE-1 suggest that it may be possible to have ‘side-on’ or asymmetric docking modes of receptor molecules. The high affinity Fab molecule binds to a relatively flat antigenic surface of the pHLA A1 where one side chain SerP8 points into the solvent and ProP4 and ThrP5 appear to be less exposed. Although the Fab/HLA A1 binding interface SC values are similar to those of the 1G4 TCR complexes, at around 0.70–0.75, the peptide on its own is poorly matched to the Fab and gives reduced SC values around 0.48–0.60 (Table IV). In contrast, the corresponding residues (ProP4 and ThrP5) appear to be well matched to the Fab with SC values around 0.72–0.92. This indicates that the Fab/peptide binding is sub-optimal, only extending to part of the peptide. Since there is more affinity to the HLA A1 helices than the peptide, there is an increased possibility of cross-reactivity with other HLAs.
We have solved the crystal structure of another in vitro affinity enhanced TCR which binds to human T-cell lymphotrophic virus type 1(HTLV-1) tax11–19 peptide. Like the high affinity 1G4 described in this paper, this high affinity TCR bound pMHC in a very similar geometry compared to its wild type parent molecule (J.M.Boulter, personal communication). In this case, the high affinity mutant was generated by a subtle opening of the CDR3ß loop to form a ‘tight pocket’ for the TyrP5 peptide side chain (to be published). This altered conformation of TyrP5 in the high affinity mutant complex relieves steric interactions observed in the native wild-type complex structure reported earlier (Garboczi et al., 1996). We propose that improved CDR3 loop/peptide contacts are a general structural feature of high affinity TCRs.
Previously, we have proposed that improved contacts at the pMHC interface may alone account for improved affinity (Dunn et al., 2006). The present study suggests that dramatic affinity gain arises via a combination of improved interactions (H-bonds, ionic associations), increased interface area and increased SC.
|
Footnotes |
|---|
Edited by David Ollis
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We thank Neil Watson for help with fermentation and David Cole, Martin Green for helpful advice and comments. Jonathan Boulter is supported by an RCUK academic fellowship.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Ashfield R., Jakobsen B.K. IDrug (2006) 9:554–559.
Boulter J.M., Glick M., Todorov P.T., Baston E., Sami M., Rizkallah P., Jakobsen B.K. Protein Eng. (2003) 16:707–711.
CCP4 (Collaborative Computational Project Number 4). Acta Crystallogr. D Biol. Crystallogr. (1994) 50:760–763.[CrossRef][Medline]
Chen J.L., et al. J. Exp. Med. (2005) 201:1243–1255.
Davis M.M., Boniface J.J., Reich Z., Lyons D., Hampl J., Arden B., Chen Y. Annu. Rev. Immunol. (1998) 16:523–544.[CrossRef][Web of Science][Medline]
Delano W.L. (2002) http://www.pymol.org.
Dunn S.M., Rizkallah P.J., Baston E., Mahon T., Cameron B., Moysey R., Gao F., Sami M., Li Y., Jakobsen B.K. Protein Sci. (2006) 15:710–721.[CrossRef][Web of Science][Medline]
Emsley P., Cowtan K. Acta Crystallography D Biol. Crystallogr. (2004) 60:2126–2132.[CrossRef]
Evans P.R. (1993) Proceedings of CCP4 Study Weekend on Data Collection & Processing. 114–122.
French G.S., Wilson K.S. Acta. Cryst. (1978) A34:517.
Garboczi D.N., Ghosh P., Utz U., Fan Q.R., Biddision W.E., Wiley D.C. Nature (1996) 384:134.[CrossRef][Medline]
Hülsmeyer M., Chames P., Hillig R.C., Stanfield R.L., Held G., Coulie P.G., Alings C., Wille G., Saenger W., Uchanska-Ziegler B., Hoogenboom H.R., Ziegler A. J. Biol. Chem. (2005) 280(4):2972–2980.
Jones T.A., Zou J.Y., Cowan S.W., Kjeldgaard M. Acta Crystallogr. (1991) A 47:110–119.[Web of Science][Medline]
Kjer-Nielsen L., Clements C.S., Purcell A.W., Brooks A.G., Whisstock J.C. Immunity (2003) 18:53–64.[CrossRef][Web of Science][Medline]
Lawrence M.C., Colman P.M. J. Mol. Biol. (1993) 234:946–950.[CrossRef][Web of Science][Medline]
Leslie A.G.W. Jnt CCP4/ESF-ECACMB Newsletter Protein Crystallography 26. (1992) MOSFM.
Li Y., et al. Nat. Biotechnol. (2005) 23:349–354.[CrossRef][Web of Science][Medline]
Murshudov G., Vagin A., Dodson E. Acta Crystallogr (1997) D53:240–255.
Molloy P.E., Sewell A.K., Jakobsen B.J. Curr. opin. Pharmacol. (2005) 5:1–6.[Medline]
Navaza J. Acta Crystallogr. D Biol. Crystallogr. (2001) 57:1367–1372.[CrossRef][Medline]
O'Callaghan C.A., Byford M.F., Wyer J.R., Willcox B.E., Jakobsen B.K., McMichael A.J., Bell J.I. Anal. Biochem. (1999) 266:9.[CrossRef][Web of Science][Medline]
Rudolph M.G., Stansfield R.L., Wilson I.A. Annu. Rev. Immunol. (2006) 24:419–466.[CrossRef][Web of Science][Medline]
Stern L.J., Wiley D.C. Structure (1994) 2:245–251.[Medline]
Tynan F.E., Borg N.A., Miles J.J., Beddoe N.A. J. Biol. Chem. (2005) 280:23900–23909.
Vagin A., Teplyakov A. J. Appl. Cryst. (1997) 30:1022–1025.[CrossRef][Web of Science]
Van der Merwe P.A., Davis S.J. Annu. Rev. Immunol. (2003) 21:659–684.[CrossRef][Web of Science][Medline]
Willcox B.E., Gao G.F., Wyer J.R., Ladbury J.E., Bell J.I., Jakobsen B.K., van der Merwe P.A. Immunity (1999a) 10:357–365.[CrossRef][Web of Science][Medline]
Willcox B.E., Gao G.F., Wyer J.R., O'Callaghan C.A., Boulter J.M., Jones E.Y., van der Merwe P.A., Bell J.I., Jakobsen B.K. Protein Sci. (1999b) 8:2418–2423.[Web of Science][Medline]
Received January 9, 2007; revised May 21, 2007; accepted June 17, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. Derre, M. Bruyninx, P. Baumgaertner, M. Ferber, D. Schmid, A. Leimgruber, V. Zoete, P. Romero, O. Michielin, D. E. Speiser, et al. Distinct sets of {alpha}{beta} TCRs confer similar recognition of tumor antigen NY-ESO-1157-165 by interacting with its central Met/Trp residues PNAS, September 30, 2008; 105(39): 15010 - 15015. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. F. Robbins, Y. F. Li, M. El-Gamil, Y. Zhao, J. A. Wargo, Z. Zheng, H. Xu, R. A. Morgan, S. A. Feldman, L. A. Johnson, et al. Single and Dual Amino Acid Substitutions in TCR CDRs Can Enhance Antigen-Specific T Cell Functions J. Immunol., May 1, 2008; 180(9): 6116 - 6131. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||